Blandford, John M. PhD; Gift, Thomas L. PhD
INFECTION WITH CHLAMYDIA TRACHOMATIS (Ct) is the most commonly reported notifiable disease in the United States and an estimated 3 million incident infections occur among sexually active adolescents and young adults annually in the United States.1–3 Although highly sensitive tests facilitate early detection of infections through routine screening, less than half of sexually active women ages 25 and younger are regularly screened.4 Because the major portion of incident Ct infections produce few or nonspecific symptoms in women, the infections may go undetected and untreated.5 As a consequence, many Ct-infected women develop serious and costly manifestations of infection.
The direct medical costs attributable to untreated Ct infection and its sequelae have been shown to be substantial.6,7 In women, untreated infection may lead to pelvic inflammatory disease (PID) and the resulting sequelae of PID, including chronic pelvic pain, ectopic pregnancy, and infertility.8 The total economic burden of a disease on society, however, includes both direct and indirect costs. The indirect costs of disease, comprising the workplace productivity losses attributable to morbidity and mortality, impose a notable burden on society in general and on employers in particular. A reliable assessment of the per-case productivity losses associated with untreated Ct infection and its sequelae would complement earlier work on direct medical costs and could contribute to a more complete estimate of the full cost of disease. This estimate may inform future evaluations of the cost-effectiveness of programs and services for the prevention, early diagnosis, and treatment of chlamydial infections. Using disease surveillance and labor market data, we modeled outcomes of untreated Ct infection in women to estimate the per-case lifetime discounted productivity losses among women attributable to untreated Ct and its sequelae.
We developed a decision-analytic model to estimate the number of lost work days among reproductive-aged women attributable to untreated lower genital tract Ct infection and to estimate the value of productivity losses associated with the lost labor time. To accommodate the large variations in Ct incidence across age, we matched age-based risk factors for Ct infection and its sequelae with age-specific labor market characteristics. We limited the study to productivity losses occurring in the paid labor force and did not consider the productivity losses among those engaged in domestic or other unpaid labor activities. This focus arose from our desire to highlight the economic burden of Ct infections to employers as well as the greater difficulty and introduced error associated with estimating the value of unpaid labor.
We estimated productivity for women in each age group using the human capital approach to valuing productivity.9,10 This widely used approach proposes that, on average, an employee's full wage, which includes the employer's share of payroll taxes and fringe benefits (here estimated at 22.44% of money wage),10 provides a reasonable approximation of the employee's marginal contribution to the firm's total output of goods and services. In essence, this approach asserts that, in a state of market equilibrium, an employee's wages will reflect his or her net additional contribution to the firm's production.
To estimate wage and labor force participation rates for each age group of reproductive-aged women, we used the 2001 Current Population Survey (CPS)–Annual Demographic Supplement compiled by the federal Bureau of Labor Statistics and Census Bureau.11 The CPS is a monthly national probability sample of approximately 56,000 households on their labor force characteristics conducted through in-person and phone interviews. The Annual Demographic Supplement of the CPS collects a full range of demographic and labor force information from the same households that participate in the monthly surveys, including data on household composition, earnings, occupation, industry, and healthcare coverage.12 For each of the age groups represented among reproductive-aged women, we calculated the distribution of full hourly wages and of usual hours worked and mean labor force participation rates (the proportion of working age women who are currently employed or actively seeking employment) (Table 1).
The most nuanced estimates of lost productivity would use actual earnings data for affected women, but these data are not publicly available. As a consequence, average societal wages are also often used to estimate productivity losses. The latter method has the advantage of encouraging social equity in the distribution of health services and other resources, because lost productivity resulting from morbidity or mortality is valued equally across persons.9 This method of estimating wages, however, produces biased estimates of lost productivity in cases in which disease incidence is not distributed uniformly across populations. Because the distribution of chlamydial infections is highly dependent on age, the method would be expected to produce highly biased indirect cost estimates. We felt a reasonable approach to estimating productivity losses attributable to untreated Ct would use wage distributions within age groups that correspond to those used for national disease surveillance.
The youngest women considered, those 15 to 19 years of age, earn low wages and have relatively low rates of labor force participation, which reflect their labor market standing as relatively low-skilled, inexperienced workers who are often still in school. Additionally, working women in this age group are likely to hold part- rather than full-time positions. On each of these labor market measures, women's average labor market standing improves as the women age, complete schooling, acquire desirable employment skills, and gain labor market experience. Therefore, each of these measures is positively correlated with age, and older women are typically remunerated at a higher rate than are younger women. By the age of 25, the means of labor force participation rates, usual hours worked, and hourly wages begin to plateau (Table 1).
Because not all women are in the paid labor force or working full-time, especially in the younger age groups, expected disability days were adjusted to reflect the likelihood that an event of PID or its sequelae would occur during a time that would interfere with work responsibilities. Thus, we calculated the expected number of lifetime disability workdays attributable to untreated Ct infection by adjusting expected workweek disability days (unspecified disability days multiplied by 5/7) by the likelihood that a women in a particular age group would be in the paid labor force (labor force participation rate) and the fraction of a full 8-hour workday that she would be expected to be on the job (usual hours worked per day/8 hours per day). For each age group of reproductive-aged women, the model multiplied daily wage rates (hourly rates × 8) by expected Ct-attributable disability workdays for each event of acute PID or its sequelae that would be expected given an untreated Ct infection.
The model incorporated the likelihood that an untreated lower genital tract Ct infection would progress to PID or the primary PID-associated sequelae of chronic pelvic pain, ectopic pregnancy pain, and tubal factor infertility. Estimates of the likelihood of progression of untreated Ct to PID and its associated sequelae were taken from published literature. Given an untreated Ct infection, we estimated a lifetime probability of 0.20 for acute PID, 0.036 for development of chronic pelvic pain, 0.012 for experiencing an ectopic pregnancy, and 0.010 for seeking and receiving infertility treatment.13,14 Based on the structure of the model, we were able to model lifetime productivity losses both from the starting point of an untreated Ct infection and from a case of PID.
For each of these adverse outcomes, we based the expected number of unspecified disability days resulting from illness and treatment on previously used estimates in the literature weighted by the likelihood that each of the conditions would be treated on an outpatient or inpatient basis (Table 2). We estimated that an episode of acute PID or chronic pelvic pain would result in 10 disability days when treated as an outpatient and 21 days when treated on an inpatient basis13–15 (Compare to unpublished data from the Pelvic Inflammatory Disease Evaluation and Clinical Health [PEACH] Randomized Trial suggests that CPP results in 6.6 additional missed work days [or 9.2 unspecified days] during a year's time. Haggerty CL. E-mail communication, September 7, 2003). For ectopic pregnancy, 28 disability days were estimated irrespective of treatment approach. Infertility treatment was estimated to result in 14 disability days and would only be offered on an outpatient basis.13–15 In each instance, we multiplied the number of unspecified disability days by 5/7 to adjust for the likelihood that the disability day fell during the workweek and could lead to a workplace productivity loss. We estimated that 89.5% of acute PID would be treated on an outpatient basis, as would 94.4% of chronic pelvic pain, 84.9% of ectopic pregnancy, and all fertility treatments.16 Because we assumed that the risk of progression from untreated Ct to PID and its sequelae was constant across age groups, the age distribution of acute PID was taken to be consistent with that of reported Ct infection rates.
Expected disability workdays and productivity losses were determined by associating patterns of disease distribution with corresponding labor market characteristics of reproductive-aged women. The Centers for Disease Control and Prevention's 2001 National Chlamydia Surveillance Data indicated that 41.2% of reported Ct infections occurring in reproductive-aged women (15–44 years of age) were among those 15 to 19 years of age, 37.5% of the infections occur among those aged 20 to 24 years, and 13.1% among women aged 25 to 29 years. Women aged 30 to 44 years make up just 8.3% of those of reproductive age with reported Ct infections.1,2 We used these surveillance data to estimate the expected distribution of incident Ct infections by age group among reproductive-aged women ages 15 to 44 years.
We assumed acute PID to occur within a year of incident Ct infection (Table 2).16 We estimated chronic pelvic pain to occur on average 2 years after infection, ectopic pregnancy 5 years after infection, and infertility treatment 10 years after infection.16 Estimated productivity losses for each event of Ct-attributable PID or its sequelae were calculated based on labor market characteristics of the woman at the age she would be expected to experience the health event. The costs of Ct-attributable sequelae of acute PID were discounted at a rate of 3% per year. All cost figures were adjusted for inflation and expressed in 2001 U.S. dollars.
We used Monte Carlo estimation techniques in the model, allowing wages and hours worked per day to vary consistent with distributions derived from the CPS data for each group. The parametric bootstrapping techniques enabled us to make best use of the rich, nationally representative set of labor market data from the CPS while allowing each of the parameters to vary simultaneously. This modeling approach provided mean estimates of expected productivity losses as well as distributional information that allowed us to calculate corresponding 95% confidence intervals.9,10 The mean estimates generated in this model closely approximate point estimates produced by standard decision-analytic approaches using fixed probabilities. For labor market parameters for which distributional data were available—comprising logged hourly wages and usual hours worked—we used a normal distribution. For labor force participation rates, generated as a mean for each age group from the reported binary measure, we assumed a uniform distribution with the range defined by the 95% exact confidence intervals. For estimates of disability days, we assumed a triangular distribution around published estimates; minimum and maximum values were set at 50% and 150% of the base values.
Within each age group and for all reproductive-aged women, we executed a Monte Carlo simulation of 1,000 hypothetical women with untreated Ct infection. The model estimated the lifetime discounted value of productivity losses attributable to an incident untreated Ct infection within each age group of women as well as for the group of all reproductive-aged women. We also performed comparable modeling of expected productivity losses expressed in terms of expected cost per case of acute PID.
The model estimated that the weighted, discounted mean lifetime productivity losses per untreated Ct infection in reproductive-aged women were $130 (confidence interval [CI] = $128–132) (Table 3). Estimated productivity losses per incident of acute PID were $649 (CI = $638–659) (Table 4). The productivity–loss estimate for PID may be compared with the discounted lifetime direct medical costs of the disease, which 2 recent studies estimated at $1,334 (2001 dollars)6 and $2,149 (2001 dollars)17 per case of PID.
We found the productivity losses attributable to untreated Ct infection and its sequelae highly correlated to age. The estimated productivity losses per Ct infection and per case of acute PID were linked to differences in mean wages, labor force participation rates, and usual hours worked. Among the youngest women, those 15 to 19 years of age, the estimated losses per untreated Ct infection were $42 (CI = $40–43); the productivity losses per case of acute PID were estimated at $208 (CI = $202–213). Attributable productivity losses occurring among women aged 20 to 24 years resulted in estimated productivity losses of $111 (CI = $109–114) per untreated Ct infection and $555 ($541–569) per case of acute PID. Among reproductive-aged women aged 25 years and older, the range of estimated productivity losses varied more narrowly. Among those women aged 25 to 44 years, estimated productivity losses per untreated Ct infection ranged from $152 to $161 and from $741 to $814 per case of acute PID. The greatest per-case expected productivity losses were estimated to occur among women aged 30 to 34 years. Women in the 30- to 34-year age group shared with older women favorable labor market characteristics (i.e., higher wages, greater participation in the paid workforce, and greater likelihood of working full-time) but were still at risk of both experiencing PID-associated ectopic pregnancy and seeking infertility treatment during their reproductive years. The estimates of PID-associated productivity losses for women aged 25 years and older are approximately 35% to 61% those of recently estimated lifetime direct medical costs6,17 and thus contribute substantially to the full burden of disease.
Assessment of the foregone productivity associated with untreated chlamydia infection and its sequelae, and the resulting impact on employers, reflects the differential earnings levels of female workers of different ages. As expected, age and wages are strongly correlated, because age serves as a marker for workplace experience, participation in the labor force, and productivity. We found that among women in the age groups with the greatest likelihood of having an untreated Ct infection and to experience acute PID and its sequelae (age groups 15–19 and 20–24), the productivity losses per untreated infection or per PID event were the least, because women of this age were less likely than older women to be in the paid labor force, to be employed in well-paying jobs, and to work full-time.
A primary limitation of the study is that the indirect cost estimates focused solely on the Ct-attributable productivity losses of working women. We did not consider the productivity losses of those women who were not in the paid labor force; neither did we include the losses that a worker would experience in productive activities outside the workplace. In both instances, we would expect substantial indirect costs associated with lost production in nonpaid activities, which include child and elder care, household management, and volunteer service. We also did not consider the productivity losses of the caregivers of those with Ct-related disability. Because of the relatively high rates of infection among young women—many of whom remain in dependent status—the productivity losses resulting from the lost labor of parents or guardians may be substantial. Each of these limitations described would, if measured, increase the indirect costs attributable to untreated chlamydial infections.
Conversely, other limitations suggest that some of the estimates of indirect costs could be overstated. The human capital approach to estimating productivity losses assumes that a worker's absence from the workplace imposes a cost to the employer either through diminished production or through the need to hire a replacement worker while providing sick leave compensation to the ill worker. Women working part-time in the labor force, however—who are disproportionately represented among the younger women—may be more likely to seek out medical care during hours and days when they are not scheduled to work. Additionally, they are less likely to be employed in positions that offer paid sick leave. In contrast, women working full-time jobs may be more likely to make use of paid sick leave to seek out medical care. As a result, the productivity losses of younger women especially could be overstated in our model.
An additional limitation arises from evolving medical management of PID and its sequelae. The projected disability days used in the model as the basis for estimated workplace productivity losses are based on previously published studies, but if improved medical management has reduced expected disability days resulting from a PID event, our estimates will proportionally overstate actual productivity losses.
In summary, our study indicates that the indirect costs of untreated Ct infections contribute substantially to the per-case economic burden of the disease. The potential to avert the significant lifetime costs associated with an untreated Ct infection—both the direct costs and productivity losses—may support efforts to expand programs to prevent Ct infections in reproductive-aged women and to increase screening, diagnosis, and treatment of Ct-infected women. Assessments of the cost-effectiveness of these public health efforts will require reasonable estimates of both the direct and indirect costs of disease.
1. Centers for Disease Control and Prevention, Division of STD Prevention. Sexually Transmitted Disease 2001 Supplement, Chlamydia Prevalence Monitoring Project. US Department of Health and Human Services, Centers for Disease Control and Prevention, December 19, 2002.
2. Centers for Disease Control and Prevention, Division of STD Prevention. Sexually Transmitted Disease Surveillance, 2001. US Department of Health and Human Services, Centers for Disease Control and Prevention, 2002.
3. Groseclose SL, Zaidi AA, DeLisle SJ, Levine WC, St. Louis ME. Estimated incidence and prevalence of genital Chlamydia trachomatis infections in the United States, 1996. Sex Transm Dis 1999; 26:339–344.
4. Johnson RE, Newhall WJ, Papp JR, et al. Screening tests to detect Chlamydia trachomatis and Neisseria gonorrhoeae infections—2002. MMWR Recomm Rep 2002; 51:1–38.
5. Stamm WE. Chlamydia trachomatis infections of the adult. In: Holmes KK, Sparling PF, Mardh P-A, Lemon SM, Stamm WE, Wasserheit JN, eds. Sexually Transmitted Disease. New York: McGraw-Hill, 1999:407–422.
6. Chesson H, Blandford J, Gift T, Tao G, Irwin K. The estimated direct medical cost of sexually transmitted diseases among American youth, 2000. Perspect Sex Reprod Health 2004; 36:11–19.
7. American Social Health Association. Sexually transmitted diseases in America: How many cases and at what cost? In: Alexander LL, Cates JR, Herndon N, Ratcliffe JF, eds. Menlo Park, CA: Kaiser Family Foundation, 1998.
8. Westrom L, Eschenbach D. Pelvic inflammatory disease. In: Holmes KK, Sparling PF, Mardh P-A, et al., eds. Sexually Transmitted Disease. New York: McGraw-Hill, 1999:783–809.
9. Drummond MF, O'Brien B, Stoddart GL, Torrance GW. Methods for the Economic Evaluation of Health Care Programmes. Oxford: Oxford University Press, 1997.
10. Haddix AC, Teutsch SM, Corso PS. Prevention Effectiveness: A Guide to Decision Analysis and Economic Evaluation. New York: Oxford University Press, 2003.
11. Current Population Survey, Annual Demographic Supplement, March 2001. US Department of Labor, Bureau of Labor Statistics and US Department of Commerce, Economic and Statistics Administration, US Census Bureau, 2001.
12. Current Population Survey Technical Paper No. 63: Design and Methodology. Washington, DC: US Department of Labor, Bureau of Labor Statistics, US Department of Commerce, Economics and Statistics Administration, US Census Bureau, 2000.
13. Marrazzo JM, Celum CL, Hillis SD, Fine D, DeLisle SJ, Handsfield HH. Performance and cost-effectiveeness of selective screening criteria for Chlamydia trachomatis infection in women. Sex Transm Dis 1997; 24:131–41.
14. van Valkengoed IG, Postma MJ, Morre SA, et al. Cost effectiveness analysis of a population based screening programme for asymptomatic Chlamydia trachomatis infections in women by means of home obtained urine specimens. Sex Transm Infect 2001; 77:276–282.
15. Washington AE, Arno PS, Brooks MA. The economic cost of pelvic inflammatory disease. JAMA 1986; 255:1735–1738.
16. Rein D, Kassler WJ, Irwin KL, Rabiee L. Direct medical cost of pelvic inflammatory disease and its sequelae: Decreasing, but still substantial. Obstet Gynecol 2000; 95:397–402.
17. Yeh JM, Hook EW, III, Goldie SJ. A refined estimate of the average lifetime cost of pelvic inflammatory disease. Sex Transm Dis 2003; 30:369–378.